BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//jEvents 2.0 for Joomla//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:STANDARD
DTSTART:20240122T104500
RDATE:20240310T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20241103T010000
RDATE:20250309T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20251102T010000
RDATE:20260308T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20261101T010000
RDATE:20270314T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20271107T010000
RDATE:20280312T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:STANDARD
DTSTART:20281105T010000
RDATE:20290311T030000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:America/New_York EST
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20240310T030000
RDATE:20241103T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20250309T030000
RDATE:20251102T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20260308T030000
RDATE:20261101T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20270314T030000
RDATE:20271107T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20280312T030000
RDATE:20281105T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:America/New_York EDT
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:cc42999fe0cf8d7a7a4d66ce7ecd8c01
CATEGORIES:Mathematical Physics Seminar
CREATED:20250102T112032
SUMMARY:Webinar: Eric Vanden-Eijnden -  Generative modeling with flows and diffusions
LOCATION:Zoom
DESCRIPTION:Eric Vanden-Eijnden – NYU\n \nWednesday, January 22nd , 10:45AM EST\n \nGen
 erative modeling with flows and diffusions\n \nGenerative models based on d
 ynamical transport have recently led to significant advances in unsupervise
 d learning. At mathematical level, these models are primarily designed arou
 nd the construction of a map between two probability distributions that tra
 nsform samples from the first into samples from the second.  While these me
 thods were first introduced in the context of image generation, they have f
 ound a wide range of applications, including in scientific computing where 
 they offer interesting ways to reconsider complex problems once thought int
 ractable because of the curse of dimensionality. In this talk, I will discu
 ss the mathematical underpinning of generative models based on flows and di
 ffusions, and show how a better understanding of their inner workings can h
 elp improve their design. These results indicate how to structure the trans
 port to best reach complex target distributions while maintaining computati
 onal efficiency, both at learning and sampling stages.  I will also discuss
  applications of generative AI in scientific computing, in particular in th
 e context of Monte Carlo sampling, with applications to the statistical mec
 hanics and Bayesian inference, as well as probabilistic forecasting, with a
 pplication to fluid dynamics and atmosphere/ocean science.\n
X-ALT-DESC;FMTTYPE=text/html:<p style="text-align: center;"><strong>Eric Vanden-Eijnden – NYU</strong></
 p><p style="text-align: center;"><strong>&nbsp;</strong></p><p style="text-
 align: center;"><strong>Wednesday,&nbsp;January 22nd ,&nbsp;10:45AM EST</st
 rong></p><p style="text-align: center;"><strong>&nbsp;</strong></p><p style
 ="text-align: center;"><strong>Generative&nbsp;modeling&nbsp;with flows and
  diffusions</strong></p><p style="text-align: center;"><strong>&nbsp;</stro
 ng></p><p>Generative&nbsp;models&nbsp;based on dynamical transport have rec
 ently led to significant advances in unsupervised learning. At mathematical
  level, these&nbsp;models&nbsp;are primarily designed around the constructi
 on of a map between two probability distributions that transform samples fr
 om the first into samples from the second.&nbsp; While these methods were f
 irst introduced in the context of image generation, they have found a wide 
 range of applications, including in scientific computing where they offer i
 nteresting ways to reconsider complex problems once thought intractable bec
 ause of the curse of dimensionality. In this talk, I will discuss the mathe
 matical underpinning of&nbsp;generative&nbsp;models&nbsp;based on flows and
  diffusions, and show how a better understanding of their inner workings ca
 n help improve their design. These results indicate how to structure the tr
 ansport to best reach complex target distributions while maintaining comput
 ational efficiency, both at learning and sampling stages.&nbsp; I will also
  discuss applications of&nbsp;generative&nbsp;AI in scientific computing, i
 n particular in the context of Monte Carlo sampling, with applications to t
 he statistical mechanics and Bayesian inference, as well as probabilistic f
 orecasting, with application to fluid dynamics and atmosphere/ocean science
 .</p>
DTSTAMP:20260828T041413
DTSTART;TZID=America/New_York:20250122T104500
DTEND;TZID=America/New_York:20250122T120000
SEQUENCE:0
TRANSP:OPAQUE
END:VEVENT
END:VCALENDAR